Human immunodeficiency virus (HIV), the causative agent of acquired immunodeficiency syndrome (AIDS), encodes three enzymes, including the well-characterized proteinase belonging to the aspartic proteinase family, the HIV protease. Inhibition of this enzyme has been regarded as a promising approach for treating AIDS. Hydroxyethylamine isosteres have been extensively utilized in the synthesis of potent and selective HIV protease inhibitors. However, this modern generation of HIV protease inhibitors has created an interesting challenge for the synthetic organic chemist. Advanced x-ray structural analysis has allowed for the design of molecules that fit closely into active sites on enzymes creating very effective drug molecules. Unfortunately, these molecules, designed by molecular shape, are often difficult to produce using conventional chemistry.
The modern generation of HIV inhibitors has structural similarities in a central three-carbon piece containing two chiral carbons that link two larger groups on each side (see, e.g., Parkes, et al., J. Org. Chem. 1994 39, 3656). Numerous synthetic routes to these isosteres have been developed. As illustrated below, a common strategy to prepare the linking group starts with an amino acid, such as phenylalanine, to set the chirality of the first carbon. Then, the linking group is completed by a series of reactions including a one-carbon homologization during which the old amino acid carbon is transformed into a hydroxy-functionalized carbon having the correct chirality. However, the commercial production of isosteres by this method presents serious challenges, generally requiring low-temperature organometallic reactions (Baragua, et al., J. Org. Chem. 1997, 62, 6080) or the use of exotic reagents. 
A second approach, which is illustrated below, is to convert the amino acid to an aldehyde and to add the carbon by use of a Wittig reaction to give an olefin (see, Luly, et al., J. Org. Chem. 1987, 52, 1487). The olefin is then epoxidized. Alternatively, the aldehyde can be reacted with nitromethane, cyanide (see, Shibata, et al., Chem. Pharm. Bull. 1998, 46, 733) or carbene sources (see, Liu, et al., Org. Proc. Res. Dev. 1997, 1, 45). Instability and difficulty in preparation of the aldehyde make these routes undesirable (see, Beaulieu, et al., J. Org. Chem. 1997, 62, 3440). 
Other routes that have been published, but not commercialized are illustrated in FIG. 1.
One of the best reagents that can be used to add a single carbon to amino acids is diazomethane because it gives high yields and few side-products. In addition, diazomethane reactions are very clean, generating only nitrogen as a by-product. HIV inhibitor molecules need high purity because of the high daily doses required. As such, diazomethane is an ideal reagent for making high purity compounds. In spite of the documented hazards of diazomethane, processes have recently been developed that permit the commercial scale use of diazomethane to convert amino acids to the homologous chloromethyl ketones (see, U.S. Pat. No. 5,817,778, which issued to Archibald, et al. on Oct. 6, 1998; and U.S. Pat. No. 5,854,405, which issued to Archibald, et al. on Dec. 29, 1998). FIG. 2 illustrates examples of HIV protease inhibitors wherein the central linking group can be synthesized by the commercial use of diazomethane. FIG. 3 illustrates a general reaction scheme that can be used to prepare the S,S-epoxide compound using diazomethane.
The most useful amino acid isosteres are based on phenylanaline. The key intermediate in the synthesis of Sequinivir® (Roche) and Aprenavir® (Glaxo Wellcome) is the (S,S-)N-t-butoxycarbonyl-1,2-epoxy-4-phenyl-3-butanamine. Several other protease inhibitors, such as those described in Chem, et al. (J. Med. Chem. 1996, 39, 1991) or those under development (e.g., BMS-234475 or BMS-232623), use the diastereomeric (R,S-)N-t-butoxycarbonyl-1,2-epoxy-4-phenyl-3-butanamine.
Beginning with readily available (L)-phenylanaline, one is able to manufacture N-t-butoxycarbonyl-1-chloro-2-keto-4-phenylbutanamine (called “chloroketone” or “CMK”) using the methods described in the literature (see, e.g., Parkes, et al., J. Org. Chem. 1994 39, 3656; Shaw, E., Methods in Enzymology (Academic Press, New York, London), 1967, 11, 677; and Dufour, et al., J. Chem. Soc. Perkin Trans. I 1986, 1895, the teachings of which are incorporated herein by reference). However, what are needed in the art are methods that allow one to produce reliably and in high-yields either diastereomer, i.e., the S,S or the R,S, from the common chloroketone starting material (see, FIG. 4). Quite surprisingly, the present invention fulfills this and other needs.